Most people treat alt text like a checkbox. They add a quick description and move on. Requirement fulfilled. 

That instinct is understandable. Alt text feels like a technical requirement tucked away in an image settings panel. It’s something you handle at the end, not something you think carefully about. Unfortunately, that mindset produces alt text that technically exists but often fails in practice, sometimes spectacularly so.

Good alt text is about communication. Getting it right requires understanding something most accessibility guides skip over: context changes everything.

What is alt text, and why is it important?

Alt text (short for alternative text) is the written description attached to an image in HTML. It looks like this: <img src=”commencement.jpg” alt=”Students in caps and gowns crossing the stage at commencement”>. When someone uses a screen reader, the alt text is what gets read aloud instead of the image. When an image fails to load, it’s the text that appears in its place. Search engines also use it to understand image content.

For higher education websites, where images do a lot of heavy lifting across academic program pages, news stories, faculty profiles, and event coverage, alt text is rarely optional. The WCAG 2.1 guidelines require meaningful alternatives for non-decorative images, and failing to provide them creates real barriers for users with visual impairments.

But compliance is the low bar, and it should be the ultimate goal. You want to give every user the same informational experience.

Why the same image can need completely different alt text

Here’s something that trips up even experienced content editors: identical images can require entirely different alt text depending on where they appear.

Take a photo of a professor standing in a biology lab. On a faculty profile page, the relevant detail is probably the person: “Professor Harper Magnusson in the biology research lab”. On a page about laboratory facilities, the lab itself might be more important: “Students working at research stations in the renovated biology lab in Anderson Hall”. On a news story about a new grant, the context shifts again.

It’s the same image, serving three different purposes. Three different alt text strings. Right?

This is why copying and pasting alt text across pages is a risky habit. The image hasn’t changed, but what the reader needs to understand has.

Stop asking “What’s in this image?” and ask instead “What would someone miss?”

The most common alt text mistake is asking the wrong question.

Asking what’s in this image leads to literal descriptions: “Photo of campus building.” While that might be technically accurate, it is practically useless.

The better question is: what would someone miss if they couldn’t see this?

That reframe shifts your focus from inventory to impact. A photo of a smiling student holding an acceptance letter on a scholarship page isn’t primarily about a student or a letter. It’s about the emotional outcome that your program delivers. An aerial shot of campus on an admissions page communicates scale, beauty, and a sense of place. A diagram in a biology course description communicates structure that no surrounding paragraph can replace.

So start by asking what matters in this specific context. Then write that.

A simple framework for deciding what to write

Before you type anything, run through these five questions:

• Does the image communicate information? If it’s purely decorative (a background texture, a stylistic divider) use alt=”” to tell screen readers to skip it.
• What would someone miss if they couldn’t see it? Identify the actual informational purpose, not just the visual content.
• Is that information already provided nearby? If the surrounding text already explains what the image shows, a brief description or empty alt may be sufficient.
• What matters in this particular context? A commencement photo on an alumni giving page serves a different purpose than the same photo in a news article about graduation rates.
• Can you say it more simply? Alt text doesn’t need to be exhaustive. Clear and concise beats thorough and dense.

Work through these in order. By the time you reach question five, you usually know exactly what to write.

Common alt text mistakes that undermine accessibility

Knowing the framework helps, but it’s equally useful to recognize what goes wrong in practice.

Writing to satisfy a requirement. Alt text that exists purely to pass an audit—“image123.jpg” or “photo”—provides no real value. A screen reader user gets nothing from it.

Keyword stuffing. Adding target keywords into alt text for SEO purposes at the expense of actual description is both a poor accessibility practice and a violation of WCAG guidance. Alt text should serve the user first.

Over-describing. Longer isn’t better. “A young woman with shoulder-length brown hair wearing a blue sweater and smiling while sitting in front of a bookshelf in what appears to be a library or study room” tells a screen reader user far more than they need for most contexts.

Assuming all images need descriptive alt text. Decorative images should have empty alt attributes (alt=””), not descriptions. Adding alt text to a purely aesthetic image forces screen reader users to sit through irrelevant information.

Using filenames. “img_4872_final_v2.jpg” is not alt text. It’s a filing system artifact that escaped into your content.

What about charts, diagrams, and infographics?

These are the cases where alt text does the heaviest lifting and where a short description is almost never enough.

A bar chart showing enrollment trends across five years on an institutional research page can’t be meaningfully summarized in a sentence. Users need the actual data. 

There are a few ways to handle this:

• Include the key insight in the alt text: “Bar chart showing undergraduate enrollment increased 18% between 2019 and 2024.”
• Provide a full data table elsewhere on the page and reference it in the alt text.
• Use ARIA aria-describedby to link to a longer text description elsewhere on the page.

The same logic applies beyond bar charts. Say your admissions office publishes a campus map showing residence halls, dining locations, and academic buildings. A prospective student using a screen reader doesn’t need every building traced in prose. They need the information the map is actually there to deliver: which residence halls sit closest to which academic buildings, and where dining options fall along that walk. Alt text like “Campus map showing three residence halls within a five minute walk of the dining hall and main academic buildings” gives them the useful part without forcing them through a verbal tour of the whole graphic.

For infographics, which are common on program comparison pages, campus maps, and research summaries, the right approach depends on complexity. A simple process diagram might be handled with a concise alt description and supporting body text. A complex infographic covering multiple data points may require a full text alternative.

The principle stays consistent: the user needs access to the same information as someone who can see the image. How you deliver that information can vary.

Can AI write alt text for you?

Yes. And that’s exactly where it gets complicated.

Modern AI tools, including built-in CMS features and image analysis APIs, can generate technically accurate descriptions of image content. Show an AI tool a photo of students gathered around a table in a campus building, and it will tell you there are people sitting around a table. That’s not wrong. But it can miss the point.

AI sees what’s in the image. It doesn’t know why you chose that image, what page it’s on, or what the surrounding content is trying to communicate. It has no way of knowing that the students in that photo are engineering majors, that the building is a newly opened facility, or that the image appears in a recruitment context where the emotional register matters as much as the factual content.

Give AI the context it needs

If you use AI to draft alt text, give it what it can’t infer on its own. Tell it:

• What page the image appears on
• What the surrounding content is about
• What purpose the image serves (emotional, informational, decorative)
• Any relevant details the AI wouldn’t know from visual analysis alone (names, departments, locations)

A prompt like “This photo appears on our mechanical engineering program page. It shows two students working together in the fabrication lab. Write alt text that emphasizes collaboration and hands-on learning” will produce something far more useful than “Describe this image.”

Don’t just accept the alt text AI gives you

Treat AI-generated alt text as a first draft, not a finished product. Read it against the context. Ask whether a screen reader user would get the same informational experience as someone looking at the image. Revise where necessary.

This is the same problem showing up again in a new form. AI cannot infer context, whether that AI is generating alt text or answering a prospective student’s question about your program. The fix is the same too: give the system the structure and context it needs instead of expecting it to guess. Alt text is just one small, concrete place where that principle plays out.

This is especially important for images of people. AI tools may misidentify individuals, make assumptions about demographics, or omit names that are clearly relevant (and sometimes clearly visible in the image itself).

Be especially careful with automatically generated alt text

Some CMS platforms and content workflows offer automatic alt text generation at upload. This can be a useful starting point for volume, but it requires a review process. Auto-generated alt text applied at scale without editorial oversight is one of the fastest ways to create an accessibility compliance problem that’s hard to audit and harder to fix.

If your institution uses automated alt text generation, build a review step into your content workflow, especially for high-visibility pages like program pages, homepages, and admissions content.

A quick alt text checklist

Before publishing any image, run through this:

• Is this image decorative? If yes, use alt=””.
• Does the alt text reflect the purpose of the image on this specific page?
• Have I avoided starting with “image of” or “photo of” (screen readers already announce this)?
• Is the alt text concise (as a rough guideline, aim for under 125 characters for simple images)?
• For charts or complex visuals, is there a text alternative that provides the full data or content?
• If AI generated the alt text, have I reviewed and edited it with context in mind?
• Does the alt text avoid keyword stuffing?

Good alt text requires both sight and understanding

AI can see the image, but only you understand why it’s there.

That distinction matters more than any technical guideline. The mechanics of alt text (the HTML attribute, the character limit, the WCAG criteria) are learnable in a couple of hours. The judgment required to write alt text that actually serves your users is harder to automate, because it depends on knowing your content, your audience, and the purpose behind every image you publish.

The next time you’re adding an image to a program page, a news story, or a faculty profile, resist the reflex to describe what you see. Ask what a user would miss. I’ve watched teams spend hours perfecting a hero image and thirty seconds on its alt text, then wonder why their accessibility audit came back with findings. The image and the alt text are doing the same job for two different audiences. Give both the same care.